Kramers degeneracy theorem in nonrelativistic QED

arXiv:0809.4471v1 [math-ph] 25 Sep 2008

Kramers degeneracy theorem in nonrelativistic QED Michael Loss∗, Tadahiro Miyao†and Herbert Spohn‡ ∗ Georgia Institute of Technology, School of Mathematics, Atlanta, Georgia 30332-0160 [email protected] †‡ Zentrum Mathematik, Technische Universit¨at M¨ unchen, D-85747 Garching, Germany † [email protected], ‡[email protected] Abstract Degeneracy of the eigenvalues of the Pauli-Fierz Hamiltonian with spin 1/2 is proven by the Kramers degeneracy theorem. The Pauli-Fierz Hamiltonian at fixed total momentum is also investigated.

1

Introduction

Among the fundamental observations in physics the anomalous Zeeman effect must be high on anybody’s list. Due to the spin of the electron the degeneracy of energy levels is lifted by the interaction with an external magnetic field. On the other hand in zero external field, there is no splitting of the energy levels observed, despite the fact that electrons carry their magnetic radiation field with them. In purely physical terms the total system, i.e., the radiation included, has a time reversal symmetry which can be used to explain that energy levels must be degenerate. This is important for understanding the spectrum of atoms but also justifies the use of effective spin Hamiltonians. It is the aim of this little note to explain all this on the basis of non-relativistic QED using the Kramers degeneracy theorem. We consider an electron coupled to the quantized radiation field described what is sometimes called the Pauli-Fierz Hamiltonian. The existence of a ground state for non-relativistic QED is by now rather well understood. The reader may consult [1, 2, 6, 8, 12] for the use of various techniques. In the absence of spin the ground state can be shown to be unique, either by estimating the overlap between ground states and the Fock vacuum [2] or relying on the Perron-Frobenius theorem via the functional integral formula for the heat kernel [8]. 1

Kramers degeneracy

2

If spin is included, the ground state is no longer unique. Exact double degeneracy of the ground state was first proven by Hiroshima and Spohn [11]. They prove first that the eigenvalue is at least doubly degenerate by a perturbative argument and then show that the degeneracy cannot be more than two by calculating the overlap with the Fock vacuum as in [2]. Both calculations require that the fine structure constant (for a given ultra violet cutoff) is sufficiently small. In this note we give a simple proof of the at least double degeneracy based on the Kramers degeneracy theorem which first appeared in [14]. In view of its importance of the result and the simplicity of the argument we decided to present this argument in a separate paper. The reader will see that our proof clarifies an essential structure behind Hiroshima and Spohn’s proof. It should be remarked that not only ground states but also all eigenvectors, should they exist, are at least doubly degenerate. As far as we know, no one has applied the Kramers theorem for this purpose to nonrelativistic QED before. Our arguments also apply to N-electron system coupled to the Maxwell field provided N is odd. The Hamiltonian H of an electron interacting with the radiation field is translation invariant and hence the total R ⊕ momentum is conserved. Thus H can be written as a direct integral H = R3 H(P ) dP , where H(P ) is the Hamiltonian with a fixed total momentum P . The existence of the ground state of H(P ) is established by [3, 5, 17] under suitable conditions. Exact double degeneracy of it was shown by Hiroshima and Spohn [11], too. Their proof is a modification of the proof of the corresponding statement for H. Kramers degeneracy theorem applies to this case as well and yields the degeneracy of every eigenvalue. Recently Hiroshima gave another proof of the ground state degeneracy by using the fact that H(P ) commutes with rotations [10]. (We remark that, his method is motivated by Sasaki [15].) This argument, however, depends on the choice of the polarization vectors of the quantized radiation field. This makes the arguments somewhat complicated. Our method is free from the choice of the polarization vector. Hiroshima’s proof, however, clarifies the symmetry property of the ground states. This paper is organized as follows. In section 2 we establish an abstract framework of the Kramers degeneracy theorem. We apply this abstract theory to the Pauli-Fierz Hamiltonian with spin 1/2 in section 3 and to the Pauli-Fierz Hamiltonian at fixed total momentum in section 4. In section 5 we remark about the N-electron system coupled to the radiation field. Section 6 is devoted to discussion of Hiroshima-Spohn’s lemma. Acknowledgements This work was supported by the DFG under the grant SP181/24. M.L. would like to acknowledge partial support by NSF grant DMS-0653374.

3

Kramers degeneracy

2

Abstract theory

2.1

Reality preserving operators

Let h be a complex Hilbert space and j be an involution on h. Namely (i) j is antilinear, (ii) j 2 = 1l, the identity on h, (iii) kjxk = kxk for all x ∈ h. Let hj = {x ∈ h | jx = x}. Then hj is a real Hilbert space. A vector x in h is said to be j-real if x ∈ hj holds. A linear operator a on h is called to be reality preserving with respect to j if jdom(a) ⊆ dom(a) (equivalently jdom(a) = dom(a)) and ajx = jax for all x ∈ dom(a). We remark that a preserves reality w.r.t. j if and only if ahj ∩ dom(a) ⊆ hj holds. We denote the set of all reality preserving operators w.r.t. j by Aj (h). Basic property of Aj (h) is stated as below. Proposition 2.1 Aj (h) is a real algebra, namely, we have the following. (i) ∀a, b ∈ Aj (h) ∀α, β ∈ R, αa + βb ∈ Aj (h). (ii) ∀a, b ∈ Aj (h), ab ∈ Aj (h) provided the product is defined. Proof. This is an easy exercise. 2

2.2

Abstract Kramers degeneracy theorem

The following proposition is an abstract version of the Kramer’s degeneracy, found in the physical literature. Proposition 2.2 Let ϑ be an antiunitary operator with ϑ2 = −1l. Let H be a self-adjoint operator. Assume that H commutes with ϑ. Then each eigenvalue of H is at least doubly degenerate. Proof. Let ϕ be an eigenvector for the eigenvalue E. Since H commutes with ϑ, one sees Hϑϕ = ϑHϕ = Eϑϕ. Hence ϑϕ is also eigenvector for E. By the antiunitarity, hϑψ1 , ϑψ2 i = hψ2 , ψ1 i for all ψ1 , ψ2 . Thus, using ϑ2 = −1l, −hϕ, ϑϕi = hϑ(ϑϕ), ϑϕi = hϕ, ϑϕi. Hence hϕ, ϑϕi = 0. 2 The following lemma is a direct consequence of the functional calculus. Proposition 2.3 Assume that H and ϑ satisfy the conditions in Proposition 2.2. Let f be a real-valued measurable function on R. Then f (H) defined by the functional calculus commutes with ϑ too.

4

Kramers degeneracy

2.3

Kramers degeneracy in the system with spin 1/2

Let us consider a directsum  Hilbert space H = h ⊕ h. Let j be an involution on j 0 is an involution on H too. Let σ1 , σ2 , σ3 be the h. Then J = j ⊕ j = 0 j 2 × 2 Pauli matrices on H:       1l 0 0 −i 0 1l . , σ3 = , σ2 = σ1 = 0 −1l i 0 1l 0

We restrict our attention to the following case. Let H0 be a semibounded self-adjoint operator on H having a form   A 0 H0 = . 0 A Hence A is self-adjoint and bounded from below. We assume the following. (H.1) A ∈ Aj (h).

Clearly each eigenvalue of H0 is doubly degenerate. We consider the perturbation by a symmetric operator HI of the form:   3 X B3 B1 − iB2 HI = σ · B = σi Bj = , B1 + iB2 −B3 j=1

where each Bi , i = 1, 2, 3 is a symmetric operator on h possessing the following properties: (H.2) Each Bi is infinitesimally small with respect to A. (H.3) iBi ∈ Aj (h), i = 1, 2, 3.

Remark that (H.3) is equivalent to jBi x = −Bi jx for all x ∈ dom(Bi ), i = 1, 2, 3. The condition (H.2) guarantees the self-adjointness of the following operator H = H0 + HI . Define an antiunitary operator by ϑ = σ2 J.

(1)

Then ϑ is an antiunitary operator satisfying ϑ2 = −1l. Theorem 2.4 Under the assumptions (H.1), (H.2) and (H.3) each eigenvalue of H is at least doubly degenerate. Proof. Noting the facts ϑσi = −σi ϑ, jBi = −Bi j, i = 1, 2, 3 and the assumption (H.1), one has ϑH0 = H0 ϑ, ϑσ · B = σ · Bϑ which implies ϑH = Hϑ. Hence, by Proposition 2.2, each eigenvalue of H is at least doubly degenerate. 2

5

Kramers degeneracy

3

Pauli-Fierz Hamiltonian with spin 1/2

The Pauli-Fierz Hamiltonian is given by HPF =


2 e 1 − i∇x + eA(x) + σ · B(x) + V (x) + Hf 2 2

acting in L2 (R3 ; C2 ) ⊗ F, where F is the photon Fock space F=

X⊕

n≥0

L2 (R3 × {1, 2})⊗sn ,

h⊗s n means the n-fold symmetric tensor product of h with the convention h⊗s 0 = C. The quantized vector potential A(x) = (A1 (x), A2 (x), A3 (x)) is given by   XZ dk p ε(k, λ) eik·x a(k, λ) + e−ik·x a(k, λ)∗ , A(x) = 2(2π)3 |k| λ=1,2 |k|≤Λ where ε(k, λ) is a polarization vector which is real valued and measurable, Λ is the ultraviolet cutoff. Here a(k, λ), a(k, λ)∗ are the annihilation and creation operators which satisfy the standard commutation relations [a(k, λ), a(q, µ)∗] = δλµ δ(k − q), [a(k, λ), a(q, µ)] = 0 = [a(k, λ)∗ , a(q, µ)∗ ]. B(x) is the quantized magnetic field defined by B(x) = rotA(x) XZ =i λ=1,2

|k|≤Λ

dk p

2(2π)3 |k|

  ik·x −ik·x ∗ (k × ε(k, λ)) e a(k, λ) − e a(k, λ) .

Hf is the field energy given by Hf =

XZ

λ=1,2

R3

dk |k|a(k, λ)∗a(k, λ).

Throughout this section, we assume the following: (V) V is infinitesimally small with respect to −∆x . Then, by [7, 9], HPF is self-adjoint on dom(−∆x )∩dom(Hf ), bounded from below. Our Hilbert space L2 (R3 ; C2 ) ⊗ F is naturally identified with L2 (R3 ; F) ⊕ L2 (R3 ; F).

6

Kramers degeneracy Under this identification, HPF is understood as follows: e HPF = H0 + σ · B(x), 2   HSpinless 0 , H0 = 0 HSpinless
2 1 HSpinless = − i∇x + eA(x) + V (x) + Hf . 2 Note the following facts: (i) σ · B(x) is infinitesimally small w.r.t. H0 .

(ii) HSpinless is self-adjoint on dom(−∆x ) ∩ dom(Hf ) and bounded from below by [7, 9]. On L2 (R3 ; F), we take the following involution: jϕ =

X⊕

n≥0

ϕ(n) (−x; k1 , λ1 , . . . , kn , λn ),

x ∈ R3 , (ki , λi) ∈ R3 × {1, 2}

P (n) (x; k1 , λ1 , . . . , kn , λn ) ∈ L2 (R3 ; F). Since the annihilation operfor ϕ = ⊕ n≥0 ϕ ator a(k, λ) acts by a(k, λ)ϕ =

X⊕ √ n≥0

n + 1ϕ(n+1) (x; k, λ, k1, λ1 , . . . , kn , λn )

for ϕ ∈ L2 (R3 ; F), one has ja(k, λ) = a(k, λ)j, ja(k, λ)∗ = a(k, λ)∗ j. Namely the annihilation and creation operators are reality preserving w.r.t. j. As a consequence, we obtain j(−i∇x ) = (−i∇x )j, jA(x) = A(x)j, jB(x) = −B(x)j, jHf = Hf j, jV (x) = V (−x)j. By the above relations, one arrives at the following: Lemma 3.1 Assume that V (−x) = V (x). (i) The spinless Hamiltonian HSpinless preserves the reality w.r.t. j, equivalently HSpinless ∈ Aj (L2 (R3 ; F)). This is corresponding to (H.1).

7

Kramers degeneracy

(ii) jB(x) = −B(x)j, that is, iBi (x) ∈ Aj (L2 (R3 ; F)). This corresponds to (H.3). Thus we can apply Theorem 2.4 to obtain the following. Theorem 3.2 Let ϑ = σ2 J with J = j ⊕ j. Then ϑ is an antiunitary operator satisfying ϑ2 = −1l. Assume that (V) holds. Moreover suppose that V (x) = V (−x). Then we obtain ϑHPF = HPF ϑ. In particular, each eigenvalue of HPF is degenerate.

4

Pauli-Fierz Hamiltonian at fixed total momentum

Let us consider the Hamiltonian at fixed total momenutm
2 e 1 HPF (P ) = P − Pf + eA(0) + σ · B(0) + Hf , P ∈ R3 , 2 2

where Pf is the field momentum defined by XZ dk ka(k, λ)∗ a(k, λ). Pf = λ=1,2

R3

Our Hilbert space is C2 ⊗ F. Under the natural identification C2 ⊗F = F⊕F, our Hamiltonian is represented as HPF (P ) = H0 (P ) + σ · B(0),   HSpinless (P ) 0 H0 (P ) = , 0 HSpinless (P )
2 1 HSpinless (P ) = P − Pf + eA(0) + Hf . 2

By [10, 13], HSpinless (P ) is positive and self-adjoint on dom(Pf2 ) ∩dom(Hf ). Moreover σ · B(0) is infinitesimally small with respect to H0 (P ). We choose an involution j by X⊕ ϕ(n) (k1 , λ1 , . . . , kn , λn ), (ki , λi ) ∈ R3 × {1, 2} jϕ = n≥0

P⊕

for each ϕ = n≥0 ϕ(n) (k1 , λ1 , . . . , kn , λn ) ∈ F. Then the annihilation and cre∗ ation operators a(k, λ), a(k, reality preserving w.r.t. j again, because the P⊕λ) are (n) action of a(k, λ) on ϕ = n≥0 ϕ (k1 , λ1 , . . . , kn , λn ) ∈ F is given by X⊕ √ n + 1ϕ(n+1) (k, λ, k1, λ1 , . . . , kn , λn ). a(k, λ)ϕ = n≥0

8

Kramers degeneracy Accordingly one can easily see that jA(0) = A(0)j, jB(0) = −B(0)j, jHf = Hf j, jPf = Pf j,

which imply that HSpinless (P ) and iBi (0), i = 1, 2, 3 are in Aj (F). Thus we can apply Theorem 2.4 and obtain the following: Theorem 4.1 Let ϑ = σ2 J with J = j ⊕ j. Then ϑ is an antiunitary operator with ϑ2 = −1l. Moreover we obtain ϑHPF (P ) = HPF (P )ϑ. In particular, each eigenvalue of HPF (P ) is at least doubly degenerate.

5

N -electron system with one fixed nucleus

In this section, we remark on an N-electron system governed by the following Hamiltonian HN =

N n  X 1 j=1

+

2

X

σ

(j)

1≤i